Code Refactoring Guidelines

In software engineering, “refactoring” source code means modifying it without changing its behaviour, and is sometimes informally referred to as “cleaning it up”. Refactoring neither fixes bugs nor adds new functionality, though it might precede either activity. Rather it improves the understandability of the code and changes its internal structure and design, and removes dead code, to make it easier to comprehend, more maintainable and amenable to change. Refactoring is usually motivated by the difficulty of adding new functionality to a program or fixing a bug in it.

Code Refactoring Guidelines

  1. Big Picture
  2. Extreme Abstraction
  3. Extreme Separation
  4. Extreme Readability
  5. Interfaces
  6. Error Handling
  7. General Issues
  8. Security
  9. General Objects

Big Picture

Don’t reinvent the wheel
Look for existing solutions to problems before creating new solutions.

Think about the big picture
Decisions within a system should be congruent with the big picture.

Document your assumptions and your decisions
Keep a journal for retrospectives.

Don’t Repeat Yourself (DRY)
Every piece of knowledge must have a single, unambiguous, authoritative representation within a system.

Plan globally, develop locally
Incremental implementations should fit into a global plan.

Extreme Abstraction

Splitters can be lumped more easily than lumpers can be split
It is easier to combine two concepts than it is to separate them.

Clump data so that there is less to think about
Clumping data cuts down on the number of concepts that have to be kept in mind.

When you’re abstract, be abstract all the way
Do not describe data items using primitive data types.

Strings are more than just a string
Treat String as a primitive data type. Describe attributes with abstract data types, instead of as Strings.

Never let a constant slip into code
Use a symbolic name for all values.

To text or not to text
Use text between programs, not within programs.

If it has collection operations, make it a collection
Collections separate object usage from object storage and hide implementation of aggregate operations.

Don’t change what it is
Create new terms rather than trying to apply new meanings to current terms

Extreme Separation

Adapt a prefactoring attitude
Eliminate duplication before it occurs.

Don’t overclassify
Separate concepts into different classes based on behavior, not on data.

Place methods in classes based on what they need

If a method does not require instance data, it should not be a member of the class. Conversely, if all the method requires is the instance data, it should be a member of the class.

Honour the class maxims
Make loosely coupled cohesive classes.

Do a little job well and you may be called upon often
Methods and classes that perform specific jobs can be reused more often.

Separate policy from implementation
Keeping the what separated from the how makes the what more readable and maintainable.

Separate concerns to make smaller concerns
Split responsibilities among multiple methods and multiple classes to simplify each method and class.

Test or Production; that is the question
Place all test-only methods in a test interface.

Build flexibility for testing
Plan for flexibility in your design to allow for ease of testing.

Decouple with associations
Association classes decouple the two classes being associated.

Split interfaces

Split a single interface into multiple interfaces if multiple clients use different portions of the interface.

Do a little and pass the buck
Add proxies to interfaces to add functionality.

Business rules are a business unto themselves
Keep business rules separate from other logic.

Extreme Readability

A rose by any other name is not a rose
Create a clearly defined name for each concept in a system.

Prototypes are worth a thousand words
A picture of an interface, such as a screen, can be more powerful than just a description.

Communicate with your code
Your code should communicate its purpose and intent.

Explicitness beats implicitness
Being explicit reduces misinterpretation.

Declaration over execution
Declarative-style programming can provide flexibility without code changes.

Use the same layout to get the same layout
Use templates or scripts for classes and methods to create consistent logic.

The easiest code to debug is that which is not written
Never write functionality that already exists in usable form.

Use the client’s language
Use the client’s language in your code to make it easier to compare the logic in the code to the logic of the client.


Create interface contracts
Design with well-defined interfaces and enforce the contracts for those interfaces.

Validate, validate, validate
At each interface, validate that the input matches the contract.

Test the interface, not the implementation
Use the contract of the interface to develop the functional tests, not the implementation behind it.

Adopt and adapt
Create the interface you desire and adapt the implementation to it.

Don’t let the cold air In

With interfaces exposed to the outside world, ensure that input validation and logging occurs.

Error Handling

Decide on a strategy to deal with deviations and errors
Determine for your system what are deviations and what are errors, and how to deal with both.

Report meaningful user messages
Error messages should be reported in the context of what the user can do about the error, instead of in terms of what the underlying error is.

Never be silent
If a method encounters an error, it should report it, not remain silent.

Consider failure an expectation, not an exception
Plan how operations should respond to failures.

General Issues

Don’t speed until you know where you are going
Make the system right, before you make it fast.

The spreadsheet conundrum
Recognize when you are making the row/column decision.

Consistency is simplicity
A consistent approach to style and solutions can make code easier to maintain.

If it can’t be tested, don’t require it
Every functionality requirement, whether formally or informally stated, should have a test created for it. If you cannot test a requirement, there is no way to determine whether you have met it.

Plan for testing
Developing test strategies in advance can lead to a better design.

Figure out how to migrate before you migrate
Considering the migration path might help you discover additional considerations in other areas of the design.

Know who it Is
Determine uniqueness criteria for objects that should be unique.

Perform a retrospective after each release
Examining your design and how you created it can help in the next release.

Nothing is perfect
There is usually a better solution, but you can stop with good enough.

See what condition your condition is in
Use state-based analysis to examine object behavior.

Get something working
Create something basic before adding refinements.

Plan your logging strategy
Determine where and how you are going to log.

More is sometimes less
Use a prewritten module with more features than you currently need and adapt it to your current needs.

Be ready to import and export
Data should be available for use outside the system via a well-defined data interface.

Avoid premature generalization

Solve the specific problem before making the solution general.


If you forget security, you’re not secure
Security should not be an afterthought. Consider it during all phases of development.

Consider privacy
Systems need to be designed with privacy in mind.

General Objects

Avoid premature inheritance
Inheritance needs time to evolve.

Think interfaces, not Inheritance
Interfaces provide more fluidity in the relationships between classes.

Overloading functions can become overloading
By using unique names, functions can be more self-describing.

When in doubt, indirect
Indirection, using either methods or data, adds flexibility.

Source: Prefactoring: Extreme Abstraction, Extreme Separation, Extreme Readability



Posted in:

9 responses to “Code Refactoring Guidelines”

  1. That’s a pretty big list, but I feel it’s missing something (maybe I missed it). Ensure that tests are written for the code you’re going to refactor before you do it. That’s the only way to be sure you’re not going to change the behavior or add bugs.

  2. I feel that some of these points are a little too abstract (excuse the pun). I refer mainly to the comment about extreme abstraction :- “Splitters can be lumped more easily than lumpers can be split”. I am not sure exactly what you mean by splitters – hopefully you are not referring to methods.

  3. @Lloyd

    I believe he’s referring to object composition – i.e. creating a new class that combines the usage of several classes to perform it’s function, instead of having that class do it all itself.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: